1. Field of the Invention
The present invention relates to managing database, and more particularly, to creating and managing a network by using database. The present invention may be used for generating a network of search keyword and associated words and manages the same, to provide an associated word in response to a search keyword as inputted.
2. Description of the Related Art
Internet users generally input search keywords in the information search system such as a web search program to locate intended information. For the users' convenience, the search system may provide the users with a list of associated words including the words that are associated in their meaning to the inputted keyword. By way of example, a user may have different outcomes between when he inputs keyword ‘train’ in the web search program and when he inputs keyword ‘high-speed train’, but the outcomes may include a keyword that can bring about more desirable result. However, people sometimes do not recall a suitable keyword to input into the web search program. For example, a user who wants to find about ‘high-speed train’ may not recall the exact word, but only recalls the word ‘train’. When the user inputs the word ‘train’, the web search program may provide not only the information acquired by the word ‘train’, but also a list of associated words possibly including the word ‘high-speed train’. Accordingly, the user may try another search with the word ‘high-speed train’.
The list of related words may include words associated in their meaning to the inputted keyword, such as hypernyms, hyponyms, synonyms, related words, substituting words, etc. Further, by utilizing the etymology of the words, the words with statistical relevancy such as independent or dependent developments of the words may be included in the list. The list of associated words may additionally include words that are linked from various viewpoints. For example, to a user who inputs a keyword ‘KIM Yu-Na’, the Korean female figure skater, a list of associated words may be provided, including: ‘Suri High School’ that Kim Yu-Na went to; ‘Vancouver Winter Olympic Games/Pyeongchang Winter Olympic Games’; ‘Kiss and Cry’—the television entertainment program that featured Kim Yu-Na; ‘Asada Mao’—the rival Japanese female figure skater; ‘Air conditioner TV commercial’ that features Kim Yu-Na; or other various associated words. The user provided with the list of associated words is able to input various additional words associated with the keyword he initially inputted and therefore, provided with a broader or more exact search range.
However, it takes professionals a considerable cost and time to work on the creation of the list of associated words. Accordingly, to automate the process of creating the list of associated words, various automation methods have been proposed, including simultaneous occurrence cartography which defines association among the words based on the possibility of simultaneous occurrence, and cartography which defines words frequently appearing in each categorized document networks as the associated words. One of the methods to create a list of associated words automatically is to create a list of words including part of the keywords inputted by the user that are actually executed, and in response to an input of a new keyword to the associated word search device, to identify associated words of the new keyword and provide the same to the user. However, because these methods generally consider statistical relationship, (without considering relationship among the words in terms of meaning), problem arises that the relationship among the automatically-created associated words is sometimes not quite convincible.
To provide a list of associated words, an information search system may evaluate association between an inputted keyword and another word, select a word with high association and include the selected word in the list of associated words. The quality of the list of associated words may be evaluated based on whether the association among the inputted keyword and the words of the list of the associated words is convincible enough. Accordingly, an evaluation method is necessary, which can effectively evaluate association between an inputted keyword and another word.